Biologic Prognostic Markers in Diffuse Large B-Cell Lymphoma Patients Treated with Dose Adjusted EPOCH-R: a CALGB 50103 Correlative Science Study

Background: Biologic prognostic markers in diffuse large B-cell lymphoma (DLBCL) have focused on putative cell of origin (germinal center (GC) vs. activated B-cell (ABC)), apoptosis, and proliferation. Such markers, shown to be predictive in CHOP-treated patients, are being validated with rituximab(R)-CHOP. CALGB 50103 is a phase 2 trial of dose adjusted etoposide, vincristine, doxorubicin, cyclophosphamide, prednisone, R (DA-EPOCH-R) therapy for DLBCL. We studied CD10, BCL6, MUM1, LMO2, BCL2, and Ki67 in CALGB 50103 to determine which markers had prognostic significance.

Methods: Prospectively procured slides were stained with appropriate primary antibodies at the Pathology Coordinating Office of the CALGB (CD10, BCL6, MUM1, BCL2, Ki67) or the Cleveland Clinic (LMO2). Slides were scored independently in 10% increments by two pathologists, with a third review in case of disagreement of >20% for all stains (mean score used as final value) except Ki67, for which image analysis (IA, Aperio, Scanscope) and a visual estimate (quartiles) was used. A 30% cutoff was used for CD10, BCL6, and MUM1. 40% was used for BCL2 and 60% for Ki67. Progression-free and event-free survival (PFS, EFS) served as the endpoints.

Results: Data for at least one of the markers were available for 53 of the 75 treated patients. The median age was 56 years (range 23 – 80) and the median follow-up for the 45 patients who are still alive is 4.5 years (range 3.2–6.0). The international prognostic index (IPI) was available in 51 patients and 33 had an IPI score of 2 while 18 had scores of 3 or 4. 11 patients had progressive disease and there were 14 treatment failures. 8 patie0nts have died. Table 1 shows the significant predictors of outcome.

Table 1: Predictors of PFS and EFS in CALGB 50103

Of the cell of origin markers (CD10, BCL6, MUM1, LMO2) only CD10 negativity was associated with an inferior PFS and EFS. GC vs. non-GC phenotype (as per Hans et al Blood 2004) and LMO2 were not associated with outcome. High Ki67% by IA was also associated with inferior PFS and EFS. Similar results were seen with Ki67 visual estimates (>75% cutoff, P<.001 and P=.006 for PFS and EFS, respectively). BCL2 expression did not predict outcome in the cohort as a whole, but was associated with poor PFS and EFS (2 year survival probability 0.67 [95% CI 0.44 –0.82, P=.022], and 0.64 [95% CI 0.42 – 0.79, P=.012], respectively) in CD10 negative cases. When adjusted for IPI, Ki67 remained an independent prognostic factor for PFS (Ki67>60% hazard ratio [HR] 7.1, P=.007; IPI 3/4 HR 7.2, P=.006) and EFS (Ki67>60% HR 5.3, P=.007; IPI 3/4 HR 7.1, P=.002). The multivariate analysis with CD10 and IPI was not possible because there are no events in CD10+ cases.

Conclusions: Lack of CD10 (suggesting non-GC origin) and high Ki67 are associated with poor outcome in DA-EPOCH-R treated patients, further validating the concept of cell of origin and proliferation as predictors of outcome in DLBCL. BCL2 remains an important predictor of outcome in patients that lack the GC marker CD10, even in rituximab treated patients.